aggression replacement training (art) for … replacement training (art) for reducing antisocial...
TRANSCRIPT
Aggression Replacement Training (ART) for reducing antisocial behaviour in adolescents and adults.
C. Kaunitz, A.K. Andershed, L. Brännström, & G. Smedslund
PROTOCOL Approval date: 4 November 2010 Publication date:
2 The Campbell Collaboration | www.campbellcollaboration.org
Contact reviewer Catrine Kaunitz Department of Knowledge Based Policy and Guidance National board of Health and Welfare 106 30 Stockholm Sweden +46 75 247 37 66 [email protected] Contribution of reviewers Internal sources of support None
3 The Campbell Collaboration | www.campbellcollaboration.org
Table of contents
TABLE OF CONTENTS 3
1 BACKGROUND 4
1.1 Description of the condition 4
1.2 Description of the intervention 4
1.3 How the intervention might work 6
1.4 Why it is important to do this review 7
2 OBJECTIVE OF THE REVIEW 9
3 METHODS 10
3.1 Criteria for considering studies for this review 10
3.2 Search methods for identification of studies 11
3.3 Data collection and analysis 13
3.4 Data synthesis 18
4 REFERENCES 20
4.1 references 20
5 TABLES 25
6 SOURCES OF SUPPORT 28
6.1 Internal sources 28
6.2 External sources 28
4 The Campbell Collaboration | www.campbellcollaboration.org
1 Background
1.1 DESCRIPTION OF THE CONDITION
Antisocial behaviour manifests itself in many different forms. A broad definition
encompasses all actions that deviate significantly from established social norms. The
type of behaviours included in this definition (e.g., substance abuse, theft,
aggression) varies between countries (and many other terms are often used
(offender, conduct disorder, delinquency) for the same behavior. Consequently there
is no common metric to measure and synthesize prevalence and incidence of
antisocial behavior.
Several approaches have been used to treat youth and adults with antisocial
behaviour. Previous systematic reviews including meta-analyses suggest that
cognitive-behavioral interventions (CBT) are among the more promising
rehabilitative treatments for antisocial behaviour in youths (Lipsey et al 2001;
Landenberger & Lipsey 2005; Armelius & Andreassen 2007). Within this approach,
a number of prevention programmes have been developed to prevent and treat
antisocial behaviour. Different cognitive-behavioral approaches have also been
found to reduce recidivism with adult violent offenders, but rehabilitation programs
for adults are still novel, and few published studies examine the recidivism outcomes
of those who take part in such programs (Polaschek & Collie, 2004). Much attention
has been paid to the importance of self-control in regulating antisocial, delinquent
and criminal behavior and there have been several efforts to create techniques and
programmes to improve self-control among particularly children and adolescents
(Piquero et al 2010). Aggression Replacement Training (ART) is an example of such
a programme.
1.2 DESCRIPTION OF THE INTERVENTION
5 The Campbell Collaboration | www.campbellcollaboration.org
The ART-method is a multimodal programme originally developed for aggressive
delinquents in residential care in New York, USA. It is a structured programme that
combines the use of techniques from cognitive therapy (based on cognitive theories)
and behavioral therapy (from learning theory). The main components include anger
management, development of social skills and moral reasoning. According to the
original manual (Goldstein, Glick & Rainer 1987) ART is a 10-week, 30 –hour
intervention, administered to groups of 8-12 youths three times a week. According to
the developers (Goldstein et al 1987), aggression has an affective component, a
behavioral component and a values component. Thus, ART came to consist of
skillstreaming to teach prosocial behaviour (behavioral component), anger control
(affective component), and moral reasoning (cognitive component). It is an
educational and training approach to replace the antisocial behavior by actively
teaching the desirable behaviours (Goldstein et al 1987). Goldstein and his colleges
encouraged the extension and modification of the program to new settings, client
groups and outcomes (Goldstein et al 2004).The programme is now available in
more or less revised forms for other forms of antisocial behaviour and populations
for example adult violent offenders. The programme was for example selected for
national implementation in English and Welsh probation service and progressed
through their offending behaviour programme accreditation prior to
implementation (Lipton et al 2000). The programme has also been implemented
and accredited in the Swedish prison and probation service (Kriminalvarden.
Retrieved, October, 27, 2010. Available at;
2http://www.kriminalvarden.se/sv/Fangelse/Arbete-klientutbildning-och-
behandling/ART/).
Skillstreaming
The behavioral component of ART consists of social skills training, a technique for
teaching pro-social behaviour to participants who are weak or lacking these
competencies. The teaching of skills serves to displace the out of control destructive
behaviours with constructive prosocial behaviour. Theoretically the method is
founded in social learning theory and the work of Bandura (1973).
The manual provides a skill checklist of 50 desired skills to identify skills that the
participants are missing, which the program then should be focusing on. Flexibility
exists to substitute some skills for others according to needs. Each skill is broken
down into its behavioral components which are modelled by the trainers and role-
played by each trainee during the session. Some skills are complicated like
understanding other people‟s feelings, while others are less complicated like
preparing yourself for a difficult conversation (Goldstein et al., 1987).
Anger management
6 The Campbell Collaboration | www.campbellcollaboration.org
The anger control training component of ART is designed to help make the
arousal of anger in chronically aggressive persons less frequent and to provide
means to learn self-control when anger is aroused. Just as skillstreaming is designed
to teach what one should do in problematic situations, anger control training
teaches what one should not do. The anger control training has its foundation in the
early anger control work of Novaco (1975) and Meichenbaum (1977).
Anger control training is a multistep sequence in which trainees are first helped to
understand how they typically perceive and interpret the behaviour of others in ways
that arouse anger. Therefore, in the first lesson, attention is given to identify the
external and internal triggers that initiate the anger. The self-control sessions
identify triggers and likely consequences of anger and aggression. The self-
awareness of triggers and arousing feelings of anger is then used to develop
alternative prosocial strategies. The trainer demonstrates the proper use of anger
reduction techniques like deep breathing and backward counting,
Moral reasoning
Moral reasoning training is the third component in ART (Gibbs et al 1995). It has
its foundation in Kohlberg‟s (1973) model of moral development with the purpose to
raise the individual‟s level of moral reasoning in order to make more mature
decisions in social situations.
In ART moral reasoning is promoted in group discussions of moral dilemmas
(social decision making meetings). Basically the trainer presents a moral dilemma
where the participants can choose between different alternatives. The trainees
choose one position each, motivate individually and discuss with one another. The
manual provides ten problem situations designed to create opportunities for
participants to consider the perspectives of others.
1.3 HOW THE INTERVENTION MIGHT WORK
Goldstein and colleagues argued that aggressive behavior and other forms of
antisocial behaviour can be traced to three factors, general shortfall in pro-social
behaviour (personal, interpersonal and social-cognitive skills), and low level of
anger control and an immature, egocentric style of moral reasoning. The skill
streaming component of ART aims to develop social skills which form pro-social
behavior. This emotion-oriented section of the programme aims to equip the
individual with self-control to manage anger and aggression. The third component
of ART addresses the concrete and egocentric thinking typically seen in those who
display aggressive behaviour. The developers claim that these components together
7 The Campbell Collaboration | www.campbellcollaboration.org
provide a programme that will help the participants to function pro-socially
(Goldstein et al 1998).
1.4 WHY IT IS IMPORTANT TO DO THIS REVIEW
A number of outcome studies of Aggression Replacement Training have been
conducted in the US (Barnoski, 2004; Coleman et al 1992) and in Europe (Hatcher
et al 2008; Gundersen & Svartdal, 2006; Moynahan, 2005; Holmqvist et al 2005).
Several studies have indicated promising results for the method for example when it
comes to recidivism. An outcome evaluation from Washington State showed for
example that, when completely delivered, ART has positive outcomes with estimated
reductions in 18-month felony recidivism of 24 percent, compared to the control
group. Most of the studies were conducted by the programme developers (Goldstein
et al 1987; Goldstein et al 1994; Leeman et al 1993).
ART-trials have been included in meta-analytic reviews of effects of a wider array of
interventions with juvenile offenders (e.g., Lipsey & Derzon 1998; Armelius et al
2007). Most of them do not report separate results for ART and no review has
specifically addressed the programme. Lipsey et al (2007) do report separate results
for ART and according to the analysis ART shows positive effects compared to
control groups when it comes to recidivism (reported OR > 5). Two independent
studies are included. Results of ART outcome studies have also been summarized
in non-systematic reviews. Several reviewers suggested that ART is a promising
empirically-based treatment for juvenile offenders (Springer et al 2003; Loeber et al
1998). Other reviews conclude that ART is an effective programme (Sherman et al
2002; Cigno & Bourn 1998). The U. S Department of Justice claim in their model
program Guide that ART is an effective program (OJJDP Model programs Guide.
Retrieved, October 27, 2010. Available at http:// www.
ojjdp.gov/mpg/mpgProgramDetails.aspx?ID=292).
Aggression Replacement Training is one of the oldest prevention programmes. Since
the 1990s it has been provided across North America and Europe within a wide
variety of social, educational, correctional services, secure units, community services
and prisons. Teachers, counselors, youth care workers, social workers, and
correctional officers are examples of people who become trainers. There have been
8 The Campbell Collaboration | www.campbellcollaboration.org
claims for the effectiveness of the intervention but few studies of ART made by
independent researchers have been included in meta-analysis and reviews despite the fact
that test searches show that several studies made by independent researchers do exist. No
review includes ART for adults. Hence, a systematic review of ART as a separate
program should be an important issue of concern for researchers, policy and for
practice.
9 The Campbell Collaboration | www.campbellcollaboration.org
2 Objective of the review
To assess the impacts of ART in residential care and community settings for
reducing antisocial behaviour in young and adult people.
10 The Campbell Collaboration | www.campbellcollaboration.org
3 Methods
3.1 CRITERIA FOR CONSIDERING STUDIES FOR THIS
REVIEW
3.1.1 Types of studies
To be included studies should be experimental, where individuals or groups are
randomly assigned to conditions, or quasi-experimental with use of parallel cohorts.
Analysis of the absolute effects of ART will involve comparing ART to no treatment
and to untreated wait list controls. The relative effects of ART (versus other
interventions) will be conducted separately and will include studies that compare
ART to other interventions and/or Treatment-As-Usual (TAU). Studies that
compare ART with ART and additional components/treatment will be excluded. All
follow-up durations reported in the primary studies will be recorded. Both
standardized and unstandardized measures will be acceptable measures.
In order to assess whether the evaluator can be regarded as independent, internet
searches will be made for each author involved in the included study. An
independent evaluator cannot have vested interests in the intervention (e.g.
economic or psychological as a developer or program proponent). In other words,
the independent evaluator should be “…free of any real or perceived bias introduced
by receipt of any benefit in cash or kind, any hospitality, or any subsidy derived from
any source that may have or be perceived to have an interest in the outcome…”
(Higgins 2008, Box 2.6a).
3.1.2 Types if participants
Participants are males and females (12 years and older) with antisocial behaviour.
Both participants in residential care (including prison, secure and open settings) and
community settings will be included. Voluntary, mandated as well as sentenced
participants will be included.
11 The Campbell Collaboration | www.campbellcollaboration.org
3.1.3 Types of interventions
Since ART has no trademark, the practical application varies. For inclusion in the
review only studies that label the programme they are evaluating “Aggression
Replacement Training” (ART) is to be included. Further, the authors have to refer to
Goldstein and include a statement that the core principles in the programme are
being followed.
3.1.4 Types of outcomes
Primary outcomes
Any recidivism in antisocial behaviour (criminal behaviour, drug-use, school
attendance) that is measured in the studies will be considered.
Secondary outcomes
Other measures of behaviour (for example social skills, interpersonal skills and
anger management), cognition (e.g., moral reasoning) and psychiatric symptoms on
standardized tests, for example Social Skills Rating System (Gresham & Elliott
1990). Measures based on unstandardized tests will only be considered if
documented psychometric properties are reported.
Analyses will be made for different follow-up periods depending on available data;
immediately post-intervention, short-term (up to one year post-intervention and
long term (over one year post intervention).
3.2 SEARCH METHODS FOR IDENTIFICATION OF STUDIES
3.2.1 Search strategy
Searches will be conducted in electronic reference databases, government databanks
and professional websites. There are no restrictions regarding language or date of
publication. To identify unpublished reports and ongoing studies, ART- developers
and independent investigators will be contacted. Reference lists of included studies
12 The Campbell Collaboration | www.campbellcollaboration.org
and all reviews will be scanned for new leads. Once a final set of included documents
is defined, this list will be sent to lead authors of included studies, together with
inclusion criteria, in order to find out if any documents are missing, including grey
literature. Conference papers (at least as titles and abstracts) are crucial when
publication bias is to be assessed. The following reference databases will be
searched:
ASSIA
Cochrane (CDSR, DARE, TRIALS, HTA)
Campbell Library
Criminal Justice Abstracts
Proquest Dissertation & Thesis
ERIC
Pub Med
PsycINFO
Sociological Abstracts
Social Work Abstracts
Social Policy and Practice (which includes Social Care Online).
Additional searches will be made by means of Google and Google Scholar and going
through the first 100 or 200 hits.
Search strategies has been developed by using various terms for aggression
replacement, aggression control therapy, aggression prevention, positive peer
culture, equipping youth to help one another, EQUIP program, Prepare curriculum,
PEACE curriculum, Family ART, etc. We will also search for studies including
descriptors/keywords describing the three components of ART; the affective
component, the behavioral component and the values component. The synonyms
from the three categories will be combined with “OR” in every category and with
“AND” between categories (see Table 1 and 2 for examples of search syntaxes for
Pub Med and PsycInfo).
It must be emphasized that syntaxes will be modified and tailored for each database
and provider. Each tailored search will include controlled terms, terms from a
thesaurus or an index (depending on database and database provider), as well as
13 The Campbell Collaboration | www.campbellcollaboration.org
free-text terms. A set of articles, that needs to be visible, will be used in order to
validate each tailored search.
3.3 DATA COLLECTION AND ANALYSIS
3.3.1 Selection of studies
Two reviewers will independently screen titles and abstracts. Selection of primary
studies will be made according to criteria described above. Studies considered
eligible by at least one of the reviewers on the basis of titles and abstracts, will be
retrieved in full text. The full texts will then be appraised by two reviewers. The same
persons will decide whether the studies meet the inclusion criteria. Any
disagreements about eligibility will be resolved by the review team. Reasons for
exclusion will be documented for each study that is retrieved in full text.
More specifically, the selection process will have the following steps (the process will
be documented by means of EndNote software and finally stored in RevMan 5.0):
1. Pairs of reviewers will independently select potentially-eligible studies for full-text
retrieval on the basis of the inclusion criteria by considering the Title, Abstract, and
Subject Terms for each document. A study will be retrieved in full text if reviewers
disagree about its potential eligibility. The results will be stored in an EndNote
database “Abstract screening”.
2. Pairs of reviewers will independently read documents in full and decide to include or
exclude the document on the basis of the inclusion criteria. If reviewers disagree, a
third reviewer will have a decisive vote. Primary reasons for exclusion will be
documented. The results will be stored in a second EndNote database “Full text
inclusion - preliminary”.
3. The complete list of included documents will be sent to a selected group of external
international experts together with the inclusion criteria. These experts will be asked
whether they know of any eligible studies that are missing studies. They will also be
asked if they know of any reasons why any of the included studies should be
excluded. Finally they will be asked if they know of any other documents (published
or not) on the included studies that could be informative during step 4.
4. Any suggestions from the external international experts will be processed in
accordance with step 2 above. The results will be stored in a third EndNote database
“Full text inclusion - final”.
14 The Campbell Collaboration | www.campbellcollaboration.org
5. The final set of included documents will be studied in order to find multiple
publications from the same study and multiple studies in single publications. The
purpose is to select a set of unique studies. The problem of multiple publications
from single studies will be approached by looking at the site and time frame of the
evaluation, the intervention, the number of participants in experiment and control
groups respectively, etc. The results will be stored in a fourth EndNote database and
in RevMan 5.0 “Included unique studies”.
3.3.2 Data extraction and management
Guided by the checklist of items to consider in data collection and data extraction
detailed in the Cochrane Handbook (Higgins 2008, Table 7.3.a), at least two
independent coders will extract data and store the data in a table in Word-format
focusing on populations, interventions, comparisons, and outcomes as basic coding
categories (see Table 3). Differences in coding will be resolved by discussion and
when not possible a third author will be adjudicated. When necessary the
corresponding author of studies will be contacted.
3.3.3 Assessment of risk of bias in included studies
Methodological quality and risk of bias regarding included studies will be assessed
independently by at least two reviewers on the basis of the revised CONSORT
statement and checklist for randomized controlled trials (Altman 2001) and the
Cochrane Handbook (Higgins 2008, section 8). Included studies will be assessed
on adequate sequence generation, allocation concealment, outcome assessors,
incomplete outcome data, selective outcome reporting, and other sources of bias, in
a risk of bias table. In all cases, an answer „Yes‟ indicates a low risk of bias, and an
answer „No‟ indicates a high risk of bias. „Unclear‟ will indicate an unclear or
unknown risk of bias. Given the nature of the method it is unlikely that providers
and participants in the intervention can be blinded; quality of blinding will be
determined primarily by whether those who assessed and coded outcome measures
were blind to treatment conditions. Since studies using quasi-experimental methods
will be included information about baseline differences and attempts made to
control for them will be examined. Details of each study will be provided in
additional tables.
15 The Campbell Collaboration | www.campbellcollaboration.org
3.3.4 Measures of treatment effect
If means and standard deviations are available continuous data will be analyzed to
calculate effect size (e.g., from t-tests, or exact p-values). Hedges g will be used for
estimating standardized mean differences (SMD) where scales measure the same
clinical outcomes in different ways (e.g., psychiatric symptoms).
If there is a mix of studies with some reporting change scores and others reporting
final values, we will contact authors and request the final values. If we do not obtain
these values, we will analyze change scores and final values separately (Higgins
2008, chapter 9, section 9.4.5.2).
For dichotomous outcomes we will calculate risk ratios (Relative Risk) with 95 %
confidence intervals and p-values. In order to calculate a common metric odds-and
risk ratios will be converted to effect size (Chinn 2000). As pointed out in Higgins
(2008, section 9.6) there are statistical approaches available which will re-express
odds/risk ratios as standardized mean differences allowing dichotomous and
continuous data to be pooled together. Even if effect sizes cannot be pooled, study-
level effects will be reported in as much detail as possible. Software for statistical
analyses will be RevMan 5.0 and STATA 10.0.
3.3.5 Unit of analysis issues
The authors will take into account the unit of analysis of the trials to determine
whether individuals were randomized in groups (i.e. cluster randomized trials),
whether individuals may have undergone multiple interventions at once, whether
results were reported at multiple time points, and whether there were multiple
treatment groups. The robustness of transforming continuous and dichotomous
outcomes into a common metric will be assessed by analyzing each outcome
measure separately.
Cluster randomized trials
It is possible that participants will be randomized to groups in clusters, either when
data from multiple participants in a setting are included (creating a cluster within
the residential or community setting), or when participants are randomized by
treatment locality or clinic. For trials that use clustered randomization, results will
be presented with proper controls for clustering (robust standard errors or
hierarchical linear models). If appropriate controls are not used and it is impossible
16 The Campbell Collaboration | www.campbellcollaboration.org
to obtain the full set of individual participant data, the data will be controlled for
clustering using the procedures outlined in Higgins (2008). That is, when outcome
measures are dichotomous, the number of events and number of participants per
trial arm will be divided by the design effect [1 + (1 - m) * r], where m is the average
cluster size and r is the intra-cluster correlation coefficient (ICC). When outcome
measures are continuous, the number of participants per trial will be divided by the
design effect, while leaving the mean values unchanged. To determine the ICC, the
reviewers will use estimates in the primary trials on a study-by-study basis.
However, where these values are not reported, the reviewers will use external
estimates of the ICC that are appropriate to each trial context and average cluster
size by contacting the trial lists and if they are not available, the reviewers will seek
statistical assistance from the Cochrane/Campbell Methods Group (Higgins 2008).
Multiple interventions per individual
If the participants in some of the trials receive ART plus treatment as usual, those
studies will be meta-analyzed separately, with the ART plus treatment as usual arm
compared to treatment as usual alone. The discussion of those results will take into
account the additional treatments received.
Multiple time points
When the results are measured at multiple time points, each outcome at each time
point will be analyzed in a separate meta-analysis with other comparable studies
taking measures at a similar time point post-intervention. These will be grouped
together as follows: immediately post-intervention, short-term (up to one year post-
intervention and long term (over one year post intervention).
Studies with multiple treatment/control groups
For trials where there are multiple treatment/control groups, data from the same
group will not be analyzed twice. Thus, multiple contrasts from the same study will
not be pooled in the same meta-analysis.
The treatment condition will be selected for meta-analysis according to which ones
match the inclusion criteria. The comparison condition will be treatment-as-usual,
the least active treatment offered, or, if neither of those is an option, the condition
that is most focused on the participant in treatment sessions.
When there is more than one control group, we will do pair-wise comparisons. This
means that effects will be calculated on the basis the following types of comparisons:
17 The Campbell Collaboration | www.campbellcollaboration.org
A vs. B, A vs. C, and B vs. C. When A is a defined intervention and C is a no-
intervention or placebo-type of intervention, we will use the term “effect”. Yet if A is
a defined intervention and B also is a defined intervention, then we will use the term
“improved effect” in comparison to the control intervention. When “usual care” is a
sufficiently described intervention, we will use the term improved effect, and if
“usual care” means no intervention “effect” will be used. In cases when “usual care”
is not defined, additional information from the authors or other reliable sources will
be used.
3.3.6 Dealing with missing data and incomplete data
Missing data and dropouts will be assessed in the included studies. Reasons,
numbers, and characteristics of dropouts will be investigated and reported. Efforts
will be made to contact the authors when further information or data are necessary.
Any meta-analyses will use data from all original participants when possible, and
will report when that is not the case. For studies in which the missing data are not
available, a sensitivity analysis will be used to assess potential bias in the analysis
and the extent to which the results might be biased by missing data will be
discussed.
Although we will seek any important but unreported data from the authors of the
original studies, it is sound to assume that this approach is not always successful. As
a consequence of this, we will also consider utilizing the imputation methods
outlined in White & Higgins (2009).
3.3.7 Assessment of heterogeneity
Heterogeneity among included studies will be examined through the use of the 2-
test, where a low p-value (p<0.1) indicates heterogeneity of treatment effects. The I2
statistic will also be used to determine the percentage of variability that is due to
heterogeneity rather than sampling error or chance (Higgins 2008). The authors will
also consider issues such as design quality, publication bias, voluntary or mandatory
participation, and differences in participant‟s characteristics as possible reasons for
any heterogeneity and conduct sensitivity analyses accordingly, where data permit.
If heterogeneity is present we will investigate possible sources using the following
steps; subgroups analyses, meta-regression and sensitivity analyses.
18 The Campbell Collaboration | www.campbellcollaboration.org
3.3.8 Assessment of reporting biases
Funnel plots will be drawn to investigate any relationship between effect size and
standard error when possible. While the visual appraisal of funnel plots will assist us
in gaining an understanding of the nature of the data, it is also appropriate with a
less subjective appraisal of the evidence for funnel plot asymmetry (Sterne & Egger,
2005). As a consequence of this, we will run a number of statistical tests for funnel
plot asymmetry (e.g. Egger‟s linear regression method).
3.4 DATA SYNTHESIS
Meta-analysis will be used when event rates or means and standard deviations are
available or can be calculated and studies include similar populations (e.g. similar
age range, criminal history, setting, etc.), methodology (e.g. randomization,
measurements, time points, etc.), and outcome measurements. Thus, when
interventions, control groups, participants, and outcomes are sufficiently similar,
pooled effects will be calculated. In other cases, when these conditions are not met,
results will be shown but not pooled (Higgins 2008, section 16.5.4). When the 2-
test or the I2 statistic indicates heterogeneity, a random effects meta-analysis will be
used. Where studies appear to be homogeneous according to known characteristics
and those statistics, a fixed effects model will be used. When meta-analysis is
inappropriate, a narrative description of the study results alone will be provided,
although general conclusions about the effectiveness of ART would not be possible
in that case.
3.4.1 Subgroup analysis, moderator analysis and investigation of
heterogeneity
Subgroup analysis and investigation of heterogeneity
Subgroups analyses will be made of ART with different subpopulations which
include:
· Service setting (juvenile justice, prisons, and open care)
· Duration of observation period
· Counterfactual condition (“usual services”, other treatment)
· Older versus younger participants.
Since previous research indicate that studies conducted by program developers
produce significantly more positive results than those conducted by independent
19 The Campbell Collaboration | www.campbellcollaboration.org
researchers (Petrosino & Soydan, 2005; Shadish et al 2002) we will also analyze
independence from program developers. We will assess results from RCTs
separately from results of quasi-experimental studies and separate analyses will
examine studies that support intent-to-treat analysis. The analyses will be conducted
to study variations in effect sizes between studies.
3.4.2 Sensitivity analysis
In order to assess the robustness of the conclusions, sensitivity analyses will be
conducted to assess the impact of the quality of the included studies. The quality
criteria used in the analyses will be as described above. The analysis will include:
· Comparing results from studies with inconsistencies in the definition,
measurement, or reporting of results (e.g. differential attrition, dropouts, lack of
intention to treat analysis, outcome measures not taken at consistent time points for
all participants) with results from consistent studies.
· Most methods for dealing with missing outcome data requires detailed data
for each participant. Since only limited information is typically available in
published reports we will focus on the case of incomplete binary outcomes (White &
Higgins 2009)
· Reanalyzing the data using different statistical approaches (e.g. using a fixed
effects model instead of a random effects model (Higgins 2008) and multivariate
meta-analysis (White 2009).
· Independence from program developers will also be considered in the
analyses (Petrosino & Soydan, 2005; Shadish et al 2002).
20 The Campbell Collaboration | www.campbellcollaboration.org
4 References
4.1 REFERENCES
Altman 2001
Altman, D. G., Schulz, K. F., Moher, D., Egger, M., Davidoff, F., Elbourne, D.,
Gøtzsche, P. C., & Lang, T. (2001). The Revised CONSORT Statement for
Reporting Randomized Trials: Explanation and Elaboration. Annals of Internal
Medicine. 134 (8), 663-694.
Armelius 2007
Armelius, B.-Å. & Andreassen, T.H. Cognitive- behavioural treatment for
antisocial behaviour in youth in residential care (Review). The Cochrane
Collaboration, John Wiley & Sons, Ltd, 2007.
Barnoski 2004
Barnoski, R. Outcome evaluation of Washington state‟s research-based
programs for juvenile offenders (Document Number: 04-01-1201). Available at:
www.wsipp.wa.gov., 2004.
Bandura 1973
Bandura, A. Aggression: A social learning analysis. Englewood Cliffs, NJ:
Prentice-Hall. 1973.
Chinn 2000
Chinn, S. A simple method for converting an odds ratio to effect size for use in
meta-analysis. Statistics in Medicine, 2000; 19:3127-3131.
Cigno 1998
Cigno, K. & Bourns, D.(eds). (1998). Cognitive-Behavioural Social Work in
Practice. Aldershot: Ashgate/Arena.
Coleman 1992
Coleman, M., Pfeiffer, S. & Oakland, T. Aggression Replacement Training with
Behaviourally Disordered Adolescents. Behavioral Disorders,1992, Vol. 18. (1), 54-
66.
21 The Campbell Collaboration | www.campbellcollaboration.org
Downs 1998
Downs SH, Black N. The feasibility of creating a checklist for the assessment of
the methodological quality both of randomised and non-randomised studies of
health care interventions. Journal of Epidemiology and Community Health 1998;
52: 377-384.
Gibbs 1995
Gibbs, J. C., Potter, G. B., & Goldstein, A. P. The EQUIP program: Teaching
youth to think and act responsibly through a peer-helping approach. Champaign,
IL: Research Press, 1995
Goldstein 1987
Goldstein, A. P, Glick, B. & Reiner, S. Aggression Replacement Training.
Champaign, IL: Research Press, 1987.
Goldstein 1994
Goldstein, A.P, Glick, B., Carthan, M., & Blancero, D. The prosocial gang;
implementing aggression replacement training. Thousand Oaks, CA: Sage, 1994.
Goldstein 1998
Goldstein, A. P., Glick, B. & Gibbs, J.C. (1998).Aggression Replacement
Training; A comprenhensive intervention for aggressive youth (2 nd ed.).
Champaign, IL: Research Press.
Goldstein 2004
Goldstein, A.P, Nensén, R., Daleflod, B. & Kalt, M. (editors). (2004). New
perspectives on aggression replacement training – Practice, Research and
Application. West Sussex; John Wiley & sons, Ltd.
Gresham 1990
Gresham, F. M. & Elliott, S. N. (1990). Social Skills Rating System. Circle Pines,
MN: American Guidance Service.
Gundersen 2006
Gundersen, K. & Svartdal, F. Aggression replacement training in Norway:
outcome evaluation of 11 Norwegian student projects. Scandinavian journal of
educational research, 2006; 50, 63-81.
Hatcher 2008
Hatcher R.M, Palmer, E., McGuire, J., Hounsome, J. Bilby, C. & Hollin, C.R.
Aggression Replacement Training with Adult Male Offenders within Community
Settings: A Reconviction Analysis. Journal of Forensic Psychiatry and psychology
(in press)
22 The Campbell Collaboration | www.campbellcollaboration.org
Higgins 2008
Higgins J.P.T &Green S. (editors). Cochrane Handbook for Systematic Reviews
of Interventions Version 5.0.2 [updated September 2009. The Cochrane
Collaboration. (available from www.cochrane-handbook.org.). 2008.
Holmqvist 2005
Holmqvist, R., Hill, T. & Lang, A. Utvärdering av behandlingen på tre
ungdomshem. Forskningsrapport nr 1, 2005. Stockholm: Statens
Institutionsstyrelse, 2005
Kohlberg 1973
Kohlberg, L. (ed). Collected papers on moral development and moral education.
Cambridge. MA: Harvard University, Center for Moral Education, 1973.
Landenberger 2005
Landenberger NA, & Lipsey MW. The positive effects of cognitive behavioral
programs for offenders: A meta-analysis of factors associated with effective
treatment. Journal of Experimental Criminology 2005;1(4):451-476.
Leeman 1993
Leeman, L.W, Gibbs, J.C & Fuller, D. (1993). Evalutation of a Multi-Component
Group Treatment Program for Juvenile Delinquents. Aggressive Behaviour, 1993;
Vol. 19. 281-292,
Lipsey 1998
Lipsey MW & Derzon JH. Predictors of violent or serious delinquency in
adolescence and early adulthood: a synthesis of longitudinal research. In: R
Loeber, D Farrington, editor(s). Serious and violent juvenile offenders: risk
factors and successful interventions. Thousand Oaks,CA. Sage Publications, 1998.
Lipsey 2001
Lipsey M.W, Chapman G.L & Landenberger N.A. Cognitive behavioral
programs for offenders. Annals of the American Academy of Political and Social
Science 2001; 578:144-4
Lipsey 2007
Lipsey M.W, Landeberger, N.A & Wilson, SJ. Effects of Cognitivebehavioral
Programs for Criminal Offenders. Campbell Systematic Reviews 2007:6.
Lipton 2000
Lipton, D., Thornton, D., McGuire, J., Porporino, F., & Hollin, C.R. Programme
accreditation and correctional treatment. Substance Use and Misuse, 2000; 35,
1705-1734.
23 The Campbell Collaboration | www.campbellcollaboration.org
Loeber 1998
Loeber, R. & Farrington, D.P (Eds) (1998). Serious and Violent Juvenile
Offenders: Risk Factors and Successful Interventions. Thousand Oaks, CA: Sage.
Meichenbaum 1977
Meichenbaum, D. H. Cognitive behaviour modification: An integrative
approach. New York: Plenum, 1977
Moynahan 2005
Moynahan, L & Stromgren, B. Preliminary results of Aggression Replacement
Training for Norwegian youth with aggressive behaviour and with a different
diagnosis. Psychology, Crime & Law, 11 (4): 411-419, 2005.
Novaco 1975
Novaco, R.W. Anger control: The development and evaluation of an
experimental treatment. Lexington, MA: D.C.Heath, 1975.
Petrosino 2005
Petrosino, A. & Soydan, H. The impact of program developers as evaluators on
criminal recidivism: Results from a meta-analysis of experimental and quasi-
experimental research, 2005. Journal of experimental criminology, 2005; 1: 435-
450.
Piquero 2010
Piquero, A.,Wesley G., Jennings, D. & Farrington, D. (2010). Self-control
interventions for children under age 10 for improving self-control and
delinquency and problem behaviours.
Campbell Systematic Reviews 2010:2.
Polaschek 2004
Polaschek, D. L. L. & Collie, R. M. (2004). Rehabilitating serious adult violent
offenders: An empirical and theoretical stocktake. Psychology, Crime and Law, 10,
321-334.
Shadish 2002
Shadish, W.R, Cook, TD & Campell D.T. Experimental and Quasi- experimental
Designs for Generalized Causal Inference. Hougton Mifflin. Boston. New York,
2002.
Sherman 2002
Sherman, L.W, Farrington, D.P, Welsh B.C & MacKenzie D.C (red) (2002).
Evidence-based Crime Prevention. London: Routledge.
24 The Campbell Collaboration | www.campbellcollaboration.org
Springer 2003
Springer, D. W., McNeece, C. A., & Arnold, E. M. (2003). Substance abuse
treatment for criminal and juvenile offenders: An evidence-based guide for
practitioners. Washington, D.C.: American Psychological Association.
Sterne 2005
Sterne, J. A. C. & Egger, M. Regression Methods to Detect Publication and
Other Bias in Meta-Analysis, pp. 99-110 in Rothstein, H. R., Sutton, A J. &
Borenstein, M. (eds.) Publication Bias in Meta-Analysis. Prevention, Assessment
and Adjustments. Chichester: John Wiley & Sons, 2005.
White 2009
White, I.R & Higgins J.P.T. Meta-analysis with missing data, pp. 218-230 in
Sterne, J. A. C. (ed.). Meta-Analysis in Stata: An updated Collection from the Stata
Journal. Stata Press, 2009.
White 2009
White, I.R. Multivariate random-effects meta-analysis, pp 231-247, in Sterne, J.
A. C. (ed.). Meta-Analysis in Stata: An updated Collection from the Stata Journal.
Stata Press, 2009.
Wells 2010
Wells GA, Shea B, O'Connell D, Peterson J, Welch V, Losos M, Tugwell P. The
Newcastle- Ottawa Scale (NOS) for assessing the quality of nonrandomised
studies in meta-analyses. Available from:
http://www.ohri.ca/programs/clinical_epidemiology/oxford.htm (accessed 25
May 2010).
25 The Campbell Collaboration | www.campbellcollaboration.org
5 Tables
Table 1: Pubmed 100317, Aggression Replacement Training (ART)
Hanna Olofsson (information specialist) & Catrine Kaunitz (lead reviewer)
Nr Search terms Numbers of ref. **)
1. aggression replacement training[tiab] OR aggression replacement therapy[tiab] OR aggression replacement treatment[tiab]
2
2. Aggression control therapy[tiab] 1
3. aggression prevention[tiab] 9
4. Positive peer culture[tiab] OR equipping youths to help one another[tiab] OR equipping youth to help one another[tiab]
0
5. prepare curriculum[tiab] OR (PREPARE AND prosocial curriculum[tiab]) OR PEACE Curriculum[tiab] OR (PEACE[tiab] AND arthur Goldstein[tw]) OR (Family ART[tiab] AND Goldstein[tw])
1
6. 1 OR 2 OR 3 OR 4 OR 5 11
7. Social Behavior[MAJR] OR Social Skills Training[tiab] OR Prosocial Behavior[tiab] OR Social Skills Training[tiab] OR Social Skills[tiab] OR prosocial skills training[tiab] OR Skillstreaming[tiab] OR self-management training[tiab]
57740
8. Anger[MAJR] OR Aggression[MAJR] OR Behavior Control[MeSH] OR Anger Control[tiab] OR anger treatment[tiab] OR Self Control[tiab] OR Aggressive Behavior[tiab] OR Anger management[tiab] OR Aggression control[tiab] OR Self-regulation Skills[tiab]
29311
9. Moral Development[MeSH] OR Morals[MAJR] OR Moral education[tiab] OR moral reasoning[tiab] OR moral judgement[tiab]
14788
10. 7 AND 8 AND 9 35
11. 6 OR 10 46
26 The Campbell Collaboration | www.campbellcollaboration.org
Table 2: PsycInfo via EBSCO 100317
Agression replacement Training (ART)
Searches by Hanna Olofsson & Catrine Kaunitz
Söknr Termtyp *) Söktermer Antal ref. **)
1. TI TI "aggression replacement" or AB "aggression replacement" or SU "aggression replacement" or MJ "aggression replacement" or DE "aggression replacement" or KW "aggression replacement" or TX "aggression replacement" or TC "aggression replacement"
55
2. TI "aggression control therapy" or AB "aggression control therapy" or SU "aggression control therapy" or MJ "aggression control therapy" or DE "aggression control therapy" or KW "aggression control therapy" or TX "aggression control therapy" or TC "aggression control therapy"
5
3. TI "aggression prevention program*" or AB "aggression prevention program*" or SU "aggression prevention program*" or MJ "aggression prevention program*" or DE "aggression prevention program*" or KW "aggression prevention program*" or TX "aggression prevention program*" or TC "aggression prevention program*"
9
4. TX "Positive peer culture" OR AB "equipping youths to help one another" OR TI EQUIP OR AB "EQUIP program" OR AB "EQUIP programme"
46
5. 1 OR 2 OR 3 OR 4 112
6. DE ”Social Skills Training" OR DE "Prosocial Behavior" OR DE Social Skills Training OR DE ”Social Skills”
14492
7. prosocial skills training OR Skillstreaming 30
8. 6 OR 7 14501
9. DE "Anger Control" or DE "Self Control" OR DE Aggressive Behavior
22735
10. ”Anger management” OR ”Anger control strategies” OR ”Anger control training” OR ”Aggression control” OR ”Self-regulation Skills”
1077
11. 9 OR 10 23321
12. DE "Moral Development" 4856
13. "Moral education" OR "moral reasoning training" OR "moral judgement"
1268
14. 12 OR 13 5518
15. 8 AND 11 AND 14 16
16. 15 OR 5 124
*)
27 The Campbell Collaboration | www.campbellcollaboration.org
Table 3: Basic categories for extracting data
Study
identification
Reviewer notes Population Interventions Control Outcomes
1) Target
population
2) Eligibility
criteria
3) Sample
description
4) Timeframe
5) Context:
Local, State,
Country
6) Other
1) Intervention
ART
2) Control 1
Usual Care, Placebo, no
intervention, etc
3) Control 2
Usual Care, Placebo, no
intervention, etc
4) Control n
Usual Care, Placebo, no
intervention, etc
5) Other
1) Randomization
procedure
2) Intervention sample
(a) Original: N1=
(b) Final: n1=
(c) Attrition at follow up
3) Control sample
(a) Original: N2=
(b) Final: n2=
(c) Attrition at follow up
4) Blinded detection
5) Adherence
6) Intention-to-treat
analysis
7) Conflicts of interest
(authors)
8) Overall
methodological quality
9) Other
1) Results and
conclusion
Results favors…
2) Primary outcomes:
measures of recidivism in
antisocial behaviour
3) Data
(a) counts and proportions
(b) means and standard
deviations
4) Calculated effect
sizes
(a) Relative Risk
(b) SMD
5) Secondary
outcomes: other
measures of behavior (e.g.
anger management)
6) Secondary
outcomes: measures of
cognition
7) Secondary
outcomes: measures of
psychiatric outcomes
8) Other
28 The Campbell Collaboration | www.campbellcollaboration.org
6 Sources of support
6.1 INTERNAL SOURCES
None
6.2 EXTERNAL SOURCES
None